Training WellGeneralizing Classifiers for Fairness Metrics and Other DataDependent Constraints
[edit]
Proceedings of the 36th International Conference on Machine Learning, PMLR 97:13971405, 2019.
Abstract
Classifiers can be trained with datadependent constraints to satisfy fairness goals, reduce churn, achieve a targeted false positive rate, or other policy goals. We study the generalization performance for such constrained optimization problems, in terms of how well the constraints are satisfied at evaluation time, given that they are satisfied at training time. To improve generalization, we frame the problem as a twoplayer game where one player optimizes the model parameters on a training dataset, and the other player enforces the constraints on an independent validation dataset. We build on recent work in twoplayer constrained optimization to show that if one uses this twodataset approach, then constraint generalization can be significantly improved. As we illustrate experimentally, this approach works not only in theory, but also in practice.
Related Material


